+++ /dev/null
-<?php
-/**
- * This file deals with UID generation.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License along
- * with this program; if not, write to the Free Software Foundation, Inc.,
- * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
- * http://www.gnu.org/copyleft/gpl.html
- *
- * @file
- * @author Aaron Schulz
- */
-
-/**
- * Class for getting statistically unique IDs
- *
- * @since 1.20
- */
-class UIDGenerator {
- /** @var UIDGenerator */
- protected static $instance = null;
-
- protected $nodeId24; // string; node ID in binary (24 bits)
- protected $nodeId32; // string; node ID in binary (32 bits)
- protected $nodeId48; // string; node ID in binary (48 bits)
- protected $threadId; // integer; thread ID or random integer
-
- protected $lockFile64; // string; local file path
- protected $lockFile88; // string; local file path
- protected $lockFile128; // string; local file path
-
- const EPOCH_MS_40 = 1325376000; // "January 1, 2012"
- const EPOCH_US_56 = 1325376000; // "January 1, 2012"
-
- protected function __construct() {
- # Try to get some ID that uniquely identifies this machine...
- wfSuppressWarnings();
- if ( wfIsWindows() ) {
- # http://technet.microsoft.com/en-us/library/bb490913.aspx
- $csv = trim( `getmac /NH /FO CSV` );
- $line = substr( $csv, 0, strcspn( $csv, "\n" ) );
- $info = str_getcsv( $line );
- $hostId = count( $info ) ? str_replace( '-', '', $info[0] ) : '';
- } else { // *nix
- # See http://linux.die.net/man/1/hostid
- $hostId = trim( `hostid` );
- }
- wfRestoreWarnings();
- $hostId = strtolower( $hostId );
- if ( preg_match( '/^[0-9a-f]{8}|[0-9a-f]{12}$/', $hostId ) ) {
- $this->nodeId24 = wfBaseConvert( substr( sha1( $hostId ), 0, 6 ), 16, 2, 24 );
- $this->nodeId32 = wfBaseConvert( substr( sha1( $hostId ), 0, 8 ), 16, 2, 32 );
- $this->nodeId48 = wfBaseConvert( $hostId, 16, 2, 48 );
- } else { // fallback to host name
- $hash = sha1( wfHostname() . ':' . php_uname( 's' ) . ':' . php_uname( 'm' ) );
- $this->nodeId24 = wfBaseConvert( substr( $hash, 0, 6 ), 16, 2, 24 );
- $this->nodeId32 = wfBaseConvert( substr( $hash, 0, 8 ), 16, 2, 32 );
- $this->nodeId48 = wfBaseConvert( substr( $hash, 0, 12 ), 16, 2, 48 );
- }
- # UID128 should be unique without having to check. The flock() calls may not be effective
- # on operating systems when PHP is running in a threaded server model. In such cases, the
- # Zend Thread Safety extension should be installed per php.net guidelines. This extension
- # gives us zend_thread_id(), which we include in the UID to avoid collisions.
- # See http://www.php.net/manual/en/install.unix.apache2.php.
- $this->threadId = function_exists( 'zend_thread_id' ) ? zend_thread_id() : 0;
- $this->lockFile64 = wfTempDir() . '/mw-' . __CLASS__ . '-UID-64.lock';
- $this->lockFile88 = wfTempDir() . '/mw-' . __CLASS__ . '-UID-88.lock';
- $this->lockFile128 = wfTempDir() . '/mw-' . __CLASS__ . '-UID-128.lock';
- }
-
- /**
- * @return UIDGenerator
- */
- protected static function singleton() {
- if ( self::$instance === null ) {
- self::$instance = new self();
- }
- return self::$instance;
- }
-
- /**
- * Get a statistically unique 64-bit unsigned integer ID string.
- * The bits of the UID are prefixed with the time (down to the millisecond).
- *
- * These IDs are suitable as values for the shard column of sharded DB tables.
- * If a column uses these as values, it should be declared UNIQUE to handle collisions.
- * New rows almost always have higher UIDs, which makes B-TREE updates on INSERT fast.
- * They can also be stored efficiently as a "BIGINT UNSIGNED" in MySQL.
- *
- * UID generation is serialized on each server (as the node ID is for the whole machine).
- *
- * @return string
- * @throws MWException
- */
- public static function timestampedUID64() {
- $generator = self::singleton();
-
- // Acquire the UID lock file
- $handle = fopen( $generator->lockFile64, 'a' );
- if ( $handle === false ) {
- throw new MWException( "Could not open '{$this->lockFile64}'." );
- } elseif ( !flock( $handle, LOCK_EX ) ) {
- fclose( $handle ); // bail out
- throw new MWException( 'Could not acquire local UID64 lock.' );
- }
- // Get the current UNIX timestamp on this server
- $time = self::millitime(); // millisecond precision
- // Delay so the next process on this server will generate higher UIDs
- while ( self::lessThanOrEqualTime( self::millitime(), $time ) );
- // Release the UID lock file
- flock( $handle, LOCK_UN );
- fclose( $handle );
-
- // Generate a UID that is higher than the last one from this server
- return $generator->newTimestampedID64( $time );
- }
-
- /**
- * @param $time array Result of UIDGenerator::millitime()
- * @return string
- */
- protected function newTimestampedID64( array $time ) {
- list( $sec, $msec ) = $time;
- $sec = $sec - self::EPOCH_MS_40; // start from epoch
- # Take the 40 MSBs of "milliseconds since epoch" (rolls over in 2046)
- if ( PHP_INT_SIZE === 8 ) { // 64 bit integers
- $ts = ( 1e3 * $sec + $msec );
- $id_bin = str_pad( decbin( $ts % pow( 2, 40 ) ), 40, '0', STR_PAD_LEFT );
- } elseif ( extension_loaded( 'bcmath' ) ) { // 32 bit integers
- $ts = bcadd( bcmul( $sec, 1e3 ), $msec );
- $id_bin = wfBaseConvert( bcmod( $ts, bcpow( 2, 40 ) ), 10, 2, 40 );
- } else {
- throw new MWException( 'bcmath extension required for 32 bit machines.' );
- }
- # Add the 24 bit node ID resulting in 64 bits total
- $id_bin .= $this->nodeId24;
- # Convert to a 1-20 digit integer string
- if ( strlen( $id_bin ) !== 64 ) {
- throw new MWException( "Detected overflow for millisecond timestamp." );
- }
- return wfBaseConvert( $id_bin, 2, 10 );
- }
-
- /**
- * Get a statistically unique 88-bit unsigned integer ID string.
- * The bits of the UID are prefixed with the time (down to the microsecond).
- *
- * These IDs are suitable as values for the shard column of sharded DB tables.
- * If a column uses these as values, it should be declared UNIQUE to handle collisions.
- * New rows almost always have higher UIDs, which makes B-TREE updates on INSERT fast.
- * They can also be stored reasonably as a "DECIMAL(27) UNSIGNED" in MySQL.
- *
- * UID generation is serialized on each server (as the node ID is for the whole machine).
- *
- * @return string
- * @throws MWException
- */
- public static function timestampedUID88() {
- $generator = self::singleton();
-
- // Acquire the UID lock file
- $handle = fopen( $generator->lockFile88, 'a' );
- if ( $handle === false ) {
- throw new MWException( "Could not open '{$this->lockFile88}'." );
- } elseif ( !flock( $handle, LOCK_EX ) ) {
- fclose( $handle ); // bail out
- throw new MWException( 'Could not acquire local UID88 lock.' );
- }
- // Get the current UNIX timestamp on this server
- $time = self::microtime(); // microsecond precision
- // Delay so the next process on this server will generate higher UIDs
- while ( self::lessThanOrEqualTime( self::microtime(), $time ) );
- // Release the UID lock file
- flock( $handle, LOCK_UN );
- fclose( $handle );
-
- // Generate a UID that is higher than the last one from this server
- return $generator->newTimestampedID88( $time );
- }
-
- /**
- * @param $time array Result of UIDGenerator::microtime()
- * @return string
- */
- protected function newTimestampedID88( array $time ) {
- list( $sec, $usec ) = $time;
- $sec = $sec - self::EPOCH_US_56; // start from epoch
- # Take the 56 MSBs of "microseconds since epoch" (rolls over in 4354)
- if ( PHP_INT_SIZE === 8 ) { // 64 bit integers
- $ts = ( 1e6 * $sec + $usec );
- $id_bin = str_pad( decbin( $ts % pow( 2, 56 ) ), 56, '0', STR_PAD_LEFT );
- } elseif ( extension_loaded( 'bcmath' ) ) { // 32 bit integers
- $ts = bcadd( bcmul( $sec, 1e6 ), $usec );
- $id_bin = wfBaseConvert( bcmod( $ts, bcpow( 2, 56 ) ), 10, 2, 56 );
- } else {
- throw new MWException( 'bcmath extension required for 32 bit machines.' );
- }
- # Add the 32 bit node ID resulting in 88 bits total
- $id_bin .= $this->nodeId32;
- # Convert to a 1-27 digit integer string
- if ( strlen( $id_bin ) !== 88 ) {
- throw new MWException( "Detected overflow for microsecond timestamp." );
- }
- return wfBaseConvert( $id_bin, 2, 10 );
- }
-
- /**
- * Get a statistically unique 128-bit unsigned integer ID string.
- * The bits of the UID are prefixed with the time (down to the microsecond).
- *
- * These IDs are suitable as globally unique IDs, without any enforced uniqueness.
- * New rows almost always have higher UIDs, which makes B-TREE updates on INSERT fast.
- * They can also be stored reasonably as a "DECIMAL(39) UNSIGNED" in MySQL.
- *
- * UID generation is serialized on each server (as the node ID is for the whole machine).
- *
- * @return string
- * @throws MWException
- */
- public static function timestampedUID128() {
- $generator = self::singleton();
-
- // Acquire the UID lock file
- $handle = fopen( $generator->lockFile128, 'a' );
- if ( $handle === false ) {
- throw new MWException( "Could not open '{$this->lockFile128}'." );
- } elseif ( !flock( $handle, LOCK_EX ) ) {
- fclose( $handle ); // bail out
- throw new MWException( 'Could not acquire local UID128 lock.' );
- }
- // Get the current UNIX timestamp on this server
- $time = self::microtime(); // microsecond precision
- // Delay so the next process on this server will generate higher UIDs
- while ( self::lessThanOrEqualTime( self::microtime(), $time ) );
- // Release the UID lock file
- flock( $handle, LOCK_UN );
- fclose( $handle );
-
- // Generate a UID that is higher than the last one from this server
- return $generator->newTimestampedID128( $time );
- }
-
- /**
- * @param $time array Result of UIDGenerator::microtime()
- * @return string
- */
- protected function newTimestampedID128( array $time ) {
- list( $sec, $usec ) = $time;
- $sec = $sec - self::EPOCH_US_56; // start from epoch
- $rand = mt_rand( 0, 255 );
- $tid = $this->threadId ? $this->threadId : mt_rand( 0, 65535 );
- # Take the 56 MSBs of "microseconds since epoch" (rolls over in 4354)
- if ( PHP_INT_SIZE === 8 ) { // 64 bit integers
- $ts = ( 1e6 * $sec + $usec );
- $id_bin = str_pad( decbin( $ts % pow( 2, 56 ) ), 56, '0', STR_PAD_LEFT );
- } elseif ( extension_loaded( 'bcmath' ) ) { // 32 bit integers
- $ts = bcadd( bcmul( $sec, 1e6 ), $usec );
- $id_bin = wfBaseConvert( bcmod( $ts, bcpow( 2, 56 ) ), 10, 2, 56 );
- } else {
- throw new MWException( 'bcmath extension required for 32 bit machines.' );
- }
- # Add on 8 bits of randomness to make 64 bits total (2^8 = 256)
- $id_bin .= str_pad( decbin( $rand ), 8, '0', STR_PAD_LEFT );
- # Add on 16 bits of thread ID or randomness to make 80 bits total (2^16 = 65536)
- $id_bin .= str_pad( decbin( $tid % 65536 ), 16, '0', STR_PAD_LEFT );
- # Add the 48 bit node ID resulting in 128 bits total
- $id_bin .= $this->nodeId48;
- # Convert to a 1-39 digit integer string
- if ( strlen( $id_bin ) !== 128 ) {
- throw new MWException( "Detected overflow for microsecond timestamp." );
- }
- return wfBaseConvert( $id_bin, 2, 10 );
- }
-
- /**
- * @return array (current time in seconds, milliseconds since then)
- */
- protected static function millitime() {
- list( $usec, $sec ) = explode( ' ', microtime() );
- return array( (int)$sec, (int)( $usec * 1e3 ) );
- }
-
- /**
- * @return array (current time in seconds, microseconds since then)
- */
- protected static function microtime() {
- list( $usec, $sec ) = explode( ' ', microtime() );
- return array( (int)$sec, (int)( $usec * 1e6 ) );
- }
-
- /**
- * Returns true if time $t1 is <= time $t2
- *
- * @param $t1 array Result of UIDGenerator::microtime() or UIDGenerator::millitime()
- * @param $t2 array Result of UIDGenerator::microtime() or UIDGenerator::millitime()
- */
- protected static function lessThanOrEqualTime( array $t1, array $t2 ) {
- return ( $t1[0] < $t2[0] || ( $t1[0] === $t2[0] && $t1[1] <= $t2[1] ) );
- }
-}